Content uploaded by Sylvia He
Author content
All content in this area was uploaded by Sylvia He on Mar 13, 2018
Content may be subject to copyright.
Mass Transit Railway, transit-oriented development, and spatial justice: The
competition for prime residential locations in Hong Kong since the 1980s
(Accepted by Town Planning Review)
Sylvia Y. He a, b, *, Sui Tao b, Yuting Hou c, Wenhua Jiang d
*
Corresponding author
a
Department of Geography and Resource Management
The Chinese University of Hong Kong
Shatin, NT, Hong Kong
Tel.: +852 3943 6646
sylviahe@cuhk.edu.hk
b
Institute of Future Cities
The Chinese University of Hong Kong
Shatin, NT, Hong Kong
c
Lee Kuan Yew Centre for Innovative Cities
Singapore University of Technology and Design
Singapore
d
Department of Civil Engineering
Monash University
Melbourne, Australia
Abstract:
Hong Kong’s Mass Transit Railway (MTR) system has been widely acclaimed for its
unique ‘rail-plus-property’ (R+P) model of operation. However, MTR projects, like other
transit-oriented development (TOD) projects, have been under scrutiny as a catalyst for
land price appreciation and gentrification. In this study, we aim to examine the impact of
new MTR lines on the spatial distribution of public and private housing estates across
Hong Kong, since the 1980s. Drawing on housing, transportation, and census data, we
reveal that local improvement in accessibility, due to the expansion of the MTR network,
has attracted private residential developments. This implies that low-income
households might have been gradually squeezed out from such prime locations. Based
on our findings, we propose recommendations on urban development under the TOD
model, with a view to making Hong Kong and other transit cities more socially and
spatially just.
Keywords:
Accessibility, Hong Kong, Private development, Public housing, Residential location,
Spatial justice, Transit-oriented development, Urban rail
3 | P a g e
Mass Transit Railway, transit-oriented development, and spatial justice: The
competition for prime residential locations in Hong Kong since the 1980s
INTRODUCTION
It has been argued that investment in transport infrastructure has a multitude of
neighbourhood impacts (Geurs et al., 2009; Won et al., 2015). Among these, the ability of
transport infrastructure to improve accessibility and stimulate local economic
development has received particular attention (Cervero, 2009; Mohammad et al., 2013).
New transport infrastructure investment improves accessibility for its nearby areas
(Cervero, 2009; Delmelle and Casas, 2012). Such improvements can cause major
changes to the neighbourhood and to the socio-demographic characteristics of its
residents (Olaru et al., 2011; Foth et al., 2013).
Stimulating the local land market through improving accessibility to other places has
been identified as a key neighbourhood effect of public transport (Duncan, 2011;
Mohammad et al., 2013). Extensive literature has linked increasing property price with
public transport infrastructure projects (e.g., Bowes and Ihlanfeldt, 2001; Du and
Mulley, 2006; Bartholomew and Ewing, 2011; Sun et al., 2015). By lifting the land and
property values in the adjacent area, public transport investment may also elicit other
neighbourhood change, particularly through gentrification, which compels the low-
income population to reside in less-costly, yet more-isolated locations (Kahn, 2007;
Revington, 2015).
Such spatial injustice is also more likely to be felt within the context of transit-oriented
development (TOD) (Knowles, 2012; Ratner and Goetz, 2013). TOD primarily involves
the concentration of high-density development of housing, employment and other
public facilities around transit stations (Cervero, 1998; Knowles, 2012; Ratner and
Goetz, 2013). Given this, it has been argued that TOD has the potential to generate an
urban form characterised by walkable, compact and mixed-use neighbourhoods, backed
with good transit connectivity (Cervero, 2006; Curtis et al., 2009). Some well-known
examples of TOD include Copenhagen (Denmark), Stockholm (Sweden), Curitiba
(Brazil), Singapore and Hong Kong (Cervero, 1998).
Hong Kong’s Mass Transit Railway (MTR) is known as one of the most successful public
transport systems in the world. In a city of over seven million people, the MTR system
enjoys a daily ridership of over 5.5 million passenger trips. This level of ridership is
among the highest across all motorised transport modes in Hong Kong (Transport
Department, 2014; MTR Corporation Limited, 2016). MTR is also one of the few transit
systems in the world that has managed to operate without relying on subsidies from the
government (Cervero, 1998; Cervero and Murakami, 2009). The success of MTR has in
large part been attributed to its unique ‘rail-plus-property’ (R+P) model. The key feature
of this business model is the pursuit of synchronised development between the rail
transit station and adjacent land (Tang et al., 2004). It integrates urban land
development into the planning and development of a rail transit service, and by doing
so, ensures stable revenue and captive ridership for the MTR (Tang et al., 2004; Cervero
and Murakami, 2009).
While widely acclaimed, Hong Kong’s MTR has also been under scrutiny in the urban
(re)development context as being a catalyst for neighbourhood gentrification. Like other
TOD projects, MTR projects operating under the R+P model have driven up the value of
adjacent land (Hong, 1998), which creates a land market that potentially encourages
4 | P a g e
more private residential development than public development (Tang et al., 2004).
Given this, the impact of MTR on the low-income community, particularly in terms of its
potential to trigger gentrification and inequitable spatial distribution of accessibility, is
subject to debate. As a backbone in Hong Kong’s public transport system, whether
MTR’s development is associated with transit-induced gentrification and the loss of
prime locations for public housing estates has never been examined. This deficit may
limit the ability to inform future MTR development proposal, which have an ambition of
achieving better social justice.
In this research, we aim to review MTR’s business model and evaluate the impact of the
R+P model on the location choice of residential development projects, whether public or
private. Drawing on census and transportation data, we analyse the longitudinal trend
and spatial distribution of private and public housing estates, in parallel with MTR
expansion, since the electrification of its operations in the late 1970s/early 1980s.
Based on the findings, a series of rail-based development recommendations for Hong
Kong is proposed. In an era where TOD is seen as a promising way forward for urban
development, in both developed and developing countries, this research, albeit focusing
on Hong Kong, may offer insights and strategies for other cities aspiring to become more
transit-oriented and socially and spatially just.
LITERATURE REVIEW
TOD, land capitalisation, and transit-induced injustice
Extensive research has shed light on the capitalisation of transport infrastructure in the
land and property market. A number of studies have focused on the influence of public
transport infrastructure (e.g., Bowes and Ihlanfeldt, 2001; Cervero and Duncan, 2002;
Du and Mulley, 2006; Sun et al., 2015). These findings in general suggest an appreciation
of property value within a walkable distance (e.g., 500 to 800 metres) from transit
services, both before and after their opening. The actual degree of appreciation,
however, varies considerably across study contexts. Through a meta-analysis of case
studies, Mohammad et al. (2013) found that this variation can be attributed to factors
related to the local land market, the degree of public transport maturation and other
social-economic contexts.
The appreciation of land and property value in the surrounding areas of TOD has been
attributed to improvements in both accessibility and the built environment
(Bartholonew and Ewing, 2011). For example, after controlling for the distance between
the rail line and stations, Mathur and Ferrell (2013) still found a positive impact of
proximity to TOD sites (i.e., within one-eighth of a mile from the sites) on the prices of
single-family homes in the City of San Jose, US. Drawing on San Diego as the case study,
Duncan (2011) highlighted that the positive effect of transit proximity on housing prices
was stronger when coupled with other TOD factors, including pedestrian-friendly
neighbourhoods, high-density living and mixed land use. Similar results have been
reported by Kay et al. (2014), where housing prices in New Jersey increase with
proximity to TOD stations.
In addition to stimulating the local land market, transport infrastructure investment
may also contribute to other important (and arguably fundamental) neighbourhood
changes, particularly through gentrification (Grube-Cavers and Patterson, 2015;
Revington, 2015). In areas that have seen increased land and property values due to
public transport improvement, two shifts in demographic composition may occur
jointly: (1) an increase in affluent households moving in and (2) an increase in lower-
income households moving out. As a result, the enhanced accessibility from transport
5 | P a g e
investment may benefit the middle-to-higher-income population more. The low-income
households, on the other hand, may experience fewer benefits from transit-oriented
development.
The gentrification associated with transport investment has been increasingly
highlighted as a threat to social justice (Grube-Cavers and Patterson, 2015). However,
few studies have shed light on such transit-induced injustice. For example, Kahn (2007)
investigated neighbourhood change within a one-kilometre distance from the light rail
transit systems across 14 US cities from 1970 to 2000. He found that in some cities (e.g.,
Boston, Washington, D.C.), communities characterised by a ‘walk and ride’ design were
more likely to gentrify than others. In another related study, Grube-Cavers and
Patterson (2015) investigated the association between the proximity to rail transit and
gentrification in the three largest Canadian cities. In this study, significant relationships
were reported for Toronto and Montreal. The findings from the above studies suggest
that transit-induced injustice has already taken place in the North American cities and
has potentially impacted the more vulnerable groups of the population by reducing
affordable housing stocks in the vicinity of transit stations. However, to what extent
transit-induced injustice exists in Asian cities where public transit plays a more
important role in facilitating daily mobility has been investigated, only to a limited
extent.
Provision and siting of public housing
The provision of public housing is a major tool to improve social equity. Given the global
rapid growth of the urban poor in the 1950s, governments around the world have
adopted various public housing policies to accommodate those most in need and
remove urban slums (Wakely and Riley, 2011; Wakely, 2014). Such policies include
constructing public housing estates, providing housing vouchers and setting up self-help
schemes. Among countries where public housing is provided, Singapore has
accommodated approximately four-fifths of its total population through public housing
development (Lee et al, 1993). Notable progress has also been made in countries such as
the US, Mexico, Brazil and Nigeria, especially in improving the living conditions for slum-
dwellers through public housing policies (UN-Habitat, 2003; Wakely, 2014).
While substantial improvement has been achieved with public housing projects around
the world, some social issues have surfaced. These include relatively lower housing
quality (Friedman, 1966; Sard and Fischer, 2008), spatial segregation of the poor and
ethnic minorities (Friedman, 1966; Oakley and Burchfield, 2009), and higher crime and
drug abuse rates (Rohe and Burby, 1988; Dunworth and Saiger, 1994). In addition to
factors such as local market forces and the poor siting of public housing plays an
important role in the proliferation of these issues. It has been noted that public housing
estates in many cases are located near the urban fringe, where land is relatively easy to
acquire. However, these areas are often associated with limited transport resources and
a paucity of necessary public facilities (Wakely, 2014). Accordingly, despite the
improved living conditions, many public housing residents face other difficulties, such as
not being able to find a job and poor access to education and other public facilities due
to inferior locations that collectively render them entrapped in their socially
disadvantaged status.
Like Singapore, public housing in Hong Kong has been adopted as a key instrument to
fulfil local accommodation needs (Chiu, 2007). The relatively low rent of public housing
has also been found to enable low-income families to undertake other meaningful
pursuits (e.g., sending kids to college or starting small businesses) (Lee and Yip, 2006).
However, to some extent mirroring the overseas experience, public housing residents in
Hong Kong may also face certain locational disadvantages for at least two main reasons.
6 | P a g e
First, given that land transactions have been a major revenue source for the Hong Kong
government, land at less-accessible locations is often assigned for public housing
development due to its lower market price (Chiu, 2007). Second, as discussed in the
previous section, Hong Kong’s TOD model appears to favour private housing
development in the precincts of MTR stations. However, the location of public housing in
relation to the MTR network expansion over the years in Hong Kong has rarely been
documented. Therefore, an examination of the spatial patterns of housing estates and
transit-based accessibility vis-à-vis MTR development in Hong Kong is timely.
BACKGROUND OF MTR AND THE R+P MODEL
Comprising a total route length of 221 km serviced by 87 main stations, the MTR serves
as a key component of Hong Kong’s public transport network and accounts for 48.5 % of
the city’s overall public transport ridership (or some 1.9 billion passenger trips on an
annual basis) (MTR Corporation Limited, 2016). It was initially owned and operated by
two companies, the MTR Corporation Limited (MTRC) and the Kowloon-Canton Railway
Corporation (KCRC). MTRC and KCRC, established in the late 1970s and 1980s,
respectively, were the two major operators of rail transit services in Hong Kong.
Operated under prudent commercial principles, the former initially provided service
routes through Hong Kong Island, while the latter supplied services between the urban
area and the New Territories (Tang and Lo, 2008). The railway constructed under the
two companies has formed the main body of today’s MTR network. In 2007, MTRC and
KCRC were merged under the name MTR to achieve enhanced operational efficiency
(Tang and Lo, 2008; Cervero and Murakami, 2009). Although previously the Hong Kong
government was the sole owner of MTR, in 2000, by offering 23% of its stock shares to
private investors, MTR transitioned into a traded company in pursuit of a more
business-minded mode of operation (Cervero and Murakami, 2009).
In a high-density city like Hong Kong, relying solely on fare revenue was still barely
sufficient to cover the construction and maintenance costs of operating a rail transit
system (Tang et al., 2004). This, in part, justified the inclusion of alternative and
arguably more-lucrative revenue sources to subsidise the MTR system (Tang et al.,
2004). Since the 1980s, an integrated Rail + Property (R+P) model has been adopted by
MTR. Three key parties are involved in the R+P model (Figure 1). In an R+P project, the
government plays an initiating role by identifying a location for railway development
and setting forth the general development strategy. Upon receiving the government’s
invitation, MTR then prepares a master plan that details the design of a new MTR station
and the development of the surrounding area in terms of density, building types and use
of urban space. If the master plan is approved for implementation, an exclusive
development right over the targeted area for a period of 50 years is granted to MTR by
the government.
<Figure 1 MTR and the R+P model>
Source: MTR (2015)
At the point of project implementation, the construction of the MTR station and
associated railway is carried out by MTR. The property development packages above
and surrounding the station are offered to private developers through public tender.
The private developers who win the contract are responsible for all the expenses related
to the construction, selling and leasing of the properties within these packages. MTR
then shifts from a plan making to a supervisory role. Finally, the management and
profits of the completed property are shared between private developers and MTR. By
7 | P a g e
the end of 2015, 96,066 residential units in conjunction with 764,018 square metres of
commercial and office space were managed by MTR (MTR Corporation Limited, 2016).
The advantages of adopting the R+P model have been shown to be numerous. First,
property-related revenue has enabled MTR not only to pay off debts and invest in the
construction of the railway but also to generate net profits to finance its rail service
(Cervero and Murakami, 2009). According to our in-depth interview (conducted in
2016) with the Head of Planning at MTR, and a review of relevant documents, property-
related revenue comes from a variety of sources, including the land premium paid by
the private developers, property rent and management fees. Such revenue has largely
fuelled the expansion of the MTR network. Currently, four new railway projects are
under implementation and are expected to add an additional 53 km of railway to the
system (MTR Corporation Limited, 2016). From a planning perspective, the synergy
between railway service and real estate properties has helped the economic benefits of
the land resources to be fully exploited, and it has guaranteed patronage since the
opening of a new MTR station (Tang et al., 2004).
Despite its successful financial performance, the R+P model has been criticised for
raising the price of nearby residential properties (Hong, 1998). Drawing on four MTR-
affiliated housing projects, Tang et al. (2004) found an average increase in housing
premium of HK$100-300 per square foot. In a related study, Cervero and Murakami
(2009) reported a 30% increase in housing prices near MTR stations. Such capitalisation
by MTR may have created a market that encourages more private than public housing
development in the precincts of MTR stations, which in turn induces social injustice. For
example, households that cannot afford or rent a private dwelling may have to opt for
more-affordable homes, yet have inferior access to MTR service. Given that the main
role of MTR is catering for mass mobility, we suspect that such injustice is taking place
across Hong Kong.
RESEARCH DESIGN AND METHODOLOGY
Research questions
Through a review of literature and background studies, we have identified that social
injustice in the housing market induced by rail transit development may exist in Hong
Kong, yet there is little related empirical evidence thus far. To bridge this gap, this study
aims to address the following research questions:
Are Hong Kong’s public housing estates located in less-desirable locations in
relation to the MTR network?
What is the impact of the MTR network on accessibility over time?
Have the locations of private and public housing estates changed as a result of
MTR network expansion?
MTR accessibility (or the accessibility to other places by MTR) was adopted as the key
indicator to capture the impacts of transit expansion. To answer the research questions,
we employed two main analytical tools, namely, time-series plots and logistic regression
models, to examine both visually and statistically the impact of MTR expansion on the
location choice of housing estates of different types.
Time-series plots
Time-series plots were used to reveal the yearly trends of the provision of public
housing vis-à-vis private housing along individual MTR lines. MTR stations were
grouped based on the lines they serve. Interchange stations serving multiple lines were
8 | P a g e
assigned to the earlier line. An examination of the opening dates of MTR lines showed
that two MTR stations, Lam Tin and LOHAS Park, were opened considerably later than
their respective lines, the Kwun Tong and Tseung Kwan O Lines. Other stations were
mostly opened within three years after the opening of a new MTR line. Due to such time
lags, they were treated as independent subjects. Apart from these two stations, a total of
12 MTR lines or line segments
1
were identified, and their opening years are summarised
in Table 1.
<Table 1 Description of MTR Lines>
To capture the quantity of public housing supply relative to its private counterpart, the
target variable was the average value after subtracting the number of private housing
units from that of public housing units. For housing estates with multiple stages, their
total housing units were divided into stage-specific numbers. If information about the
specific stage of a housing estate was not available, the housing units of a multi-stage
estate were then divided proportionally with the number of blocks built in different
years.
(1)
(2)
Where: is the difference of housing units; is the number of public
housing units; is the number of private housing units; is the
number of MTR stations associated with a particular line; and
is the average
difference of housing units.
The plotting analysis was carried out at two levels of catchment areas, i.e., 500- and
1000-metre radii. Following the example of some previous studies (Tang et al., 2004;
Cervero and Kurakami, 2009), a 500-metre Euclidean distance radiating from the
pertinent MTR station was employed as its immediate catchment area, which is the main
focus of the plotting analysis. The 1000-metre Euclidean distance was employed to
capture MTR’s broader yet still walkable areas. For each of the MTR lines, we considered
a period spanning from ten years prior to the opening of the service until 2016. For
example, for an MTR line that opened in 1995, the time range for plotting would be
1985-2016. This was to acknowledge that the influence of rail transit development on
the land market might be present during both its planning and implementation stages.
Drawing on the spatial and temporal scopes defined above, the annual
s at the two
levels of catchment areas were calculated for each of the MTR lines (including the two
stations: Lam Tin and LOHAS Park).
Regression models
Regression models were estimated to capture the relationship between the types of
housing estates (public or private) and the impact of MTR development. The dependent
variable addresses whether a major housing estate is private or not: 1 if yes, and 0 if no.
1
Six MTR stations, Airport, AisaWorldExpo, Lok Ma Chau, Sunny Bay, Disneyland Resort and
Racecourse, were excluded from this analysis. The first five stations were excluded due to their
relatively isolated locations (near the airport and the border of Hong Kong) and because they
basically have no housing estates in their respective vicinity. The Racecourse station is only
operated when there is a horserace, not for daily use; hence, it too was excluded.
9 | P a g e
Five main groups of independent variables are considered in the regression models.
First, following our previous discussion, the MTR network accessibility at the street
block level was included as a key influential factor in the type of housing estate. It is
hypothesised that a private housing estate is more likely to be built in a neighbourhood
with higher MTR accessibility. Secondly, the literature has shown that the housing
market can react to both the realised and anticipated benefits of a public transport
service (Bartholonew and Ewing, 2011). Hence, a change in MTR accessibility both
before and after a housing estate’s construction was included in the models as well. Two
periods of three years and five years were used to capture both short- and longer-term
reactions, respectively. Thirdly, dummy variables of whether an MTR station is available
within the 500-metre area from a housing estate three and five years before and after its
opening (yes = 1 and no = 0) were included to capture the presence of MTR expansion.
Fourthly, dummy variables based on time lags between the opening years of MTR lines
were employed to control for housing estates of different periods. Given the starting
point of MTR electrification and the five-year lag, we focused on estates between 1984
(five years after 1979) and 2011 (five years before 2016). Lastly, previous studies on the
impact of transit on land and property values have found that neighbourhood
characteristics, including population density, the presence of higher income households
and minority groups, may also influence the spatial distribution of housing estates (e.g.,
Mathur and Ferrell, 2013; Du and Mulley, 2006). Such neighbourhood variables were
also considered in the models.
Considering that the dependent variable is dichotomous, we employed a binary logistic
model. The basic form of the model is:
(3)
Where: Y is the probability that a housing estate is private (as opposed to public); A is
the neighbourhood MTR accessibility; ΔA is the change in neighbourhood MTR
accessibility over the past/next 3 (or 5) years; S is the dummy variable for the opening
of an MTR station in the previous/next 3 (or 5) years within the 500-metre distance
from a housing estate; T is the dummy variable for the year when the MTR line was
opened; and represents neighbourhood characteristics.
Measurement of variables
Four neighbourhood variables concerning socio-economic composition and
employment status were derived from census data at the Tertiary Planning Unit (TPU)
level. Pairwise correlations among the four potential variables were first examined. It
was found that the proportion of households with higher income is highly correlated
with the proportions of the working population and the population with higher
education (e.g., Pearson Correlation > 0.7). Hence, only household income and
population density were included in the models.
MTR accessibility was calculated across the 2011 street blocks (SB).
2
To measure this
variable, a formula used by Giuliano et al. (2012) was employed, which was used to
calculate road network accessibility. Hence, we slightly modified the original formula to
measure MTR accessibility, which takes the form:
(4)
2
The census data are not provided at the SB level. However, it is a geographic unit that is
considerably smaller than TPU. Hence, we used this unit to calculate MTR accessibility.
10 | P a g e
Where: is the MTR network accessibility of SB i to all other SBs; is the shortest
travel time between SBs i and j by MTR; and β refers to the impedance parameter to
capture the transport cost in accessing a location, which takes the value of the inverse of
average travel time
3
by public transport in Hong Kong.
Following Liu and Zhu (2004), the estimation of followed the expression:
(5)
Where: is the time of walking to the nearest MTR stations from an origin i (i.e., access
time); is the travel time of riding the MTR service between two stations nearest to i
and j (i.e., in-vehicle time); and is the time of walking to a destination j (i.e., egress
time). Both the access and egress times were calculated based on shortest path
algorithm in ArcGIS, and an average walking speed of 4.8 km per hour was assumed. The
in-vehicle time was calculated using the network analysis modules of ArcGIS based on
the average speed of different lines across the MTR network (MTR, 2017).
Considering the difference of accessibility across different years, standardised values
(ranging between -2.1 and 2.3) were used in the models. Drawing on the standardised
values, change in MTR accessibility over time was calculated last. All the accessibility-
related variables along with the neighbourhood variables are summarised in the
Appendix.
STUDY AREA AND DATA
Study area
Hong Kong has a total land area of 1,075 square kilometres, of which just over 20% is
built-up urban area. Hong Kong consists of three major regions, namely, Hong Kong
Island, Kowloon and New Territories. The former two constitute Hong Kong’s urban
area, whilst New Territories is largely the city’s suburban area. As of mid-2016, the
population of Hong Kong was over 7.3 million (Census and Statistics Department, 2016),
of which approximately 3.47 million live in the new towns in New Territories and the
rest mainly reside in the urban area (Civil Engineering and Development Department,
2016).
According to the 2011 Population Census, 51% of the population in Hong Kong lived in
private housing estates compared to 47% in public housing estates (Census and
Statistics Department, 2012).
Data
Two data sets were employed as the main information sources: housing estate data from
the Census and transport network data from Open Street Map – an open data source
that contains detailed information about the transport network of Hong Kong. The
3
Ideally, the average travel time by MTR should be used to estimate β. However, such
information is not publicly available. Therefore, we used average travel time by public transport
at four time points (i.e., 50 minutes for 1982, 42 minutes for 1992, and 43 minutes for 2002 and
2011) to estate β (Transport Department, 1993, 2003, 2014). Specifically, 50 minutes was used
to calculate accessibility prior to 1982. The average travel times between 1982 and 1992 and
between 1992 and 2002 were interpolated. The estimated time was used to calculate
accessibility in these years. Finally, 43 minutes was used to calculate the accessibility in 2000 and
after.
11 | P a g e
housing estate data stores socio-demographic and economic information (e.g.,
population, mean household income) associated with major housing estates, defined in
2011 as having at least 3,000 residents or 1,000 households (Census and Statistics
Department, 2012). This data set was further enriched to include primary information
about the estates, such as whether the estate is public or private, the intake year and the
address. Various information sources were drawn upon in this data-enrichment process.
For public housing estates, the online databases of the major public housing providers,
in particular, the Hong Kong Housing Authority and the Hong Kong Housing Society,
were referred to. For private housing estates, the online transaction records from major
estate agencies in Hong Kong, such as the Centaline Property Agency and Midland
Holdings, were used as supplementary data sources. All the housing estates were geo-
coded based on their addresses. For estates with multiple blocks, the geometric
centroids were extracted as their locations. The second data set, transport network data,
contains information that specifies (1) the configuration, length and speed limit of all the
road segments in Hong Kong; and (2) the spatial layout and operational speed of all MTR
lines and the locations of MTR stations. Figure 2 shows the locations of the major
housing estates in Hong Kong.
<Figure 2 Location of housing estates>
Descriptive analysis
According to the 2011 Census, there are a total of 487 major housing estates in Hong
Kong, of which 327 are public housing estates. Table 2 provides some basic housing and
population information for public and private housing estates. Public housing estates
have over twice as many housing units and households as private housing estates and
almost triple the population of the latter. This relationship is reversed in terms of the
average of median household monthly income, which confirms that public housing
estates indeed host more lower-income households. We next explored the proximity of
MTR stations to the housing estates. This variable was estimated as the walking time
from the housing estates to the closest MTR station using the method described
previously. A comparison of the statistics revealed modest differences in terms of the
proximity to MTR stations between public and private housing estates. Specifically, the
median walking time to the closest MTR stations from private housing estates is slightly
less than that of their public counterparts. However, the mean and population-weighted
walking times to the closest MTR stations are also larger for private housing estates,
suggesting that some of them are at relatively distant locations. In general, the statistics
above do not suggest that public housing estates are particularly disadvantaged in terms
of location. Whether this has been the case over time needs further investigation and is
addressed in the next section.
<Table 2 Descriptive statistics of housing estates>
RESULTS AND DISCUSSION
Time-series plots
Following the methods described previously, multi-stage housing estates were split
based on their year of intake. This expanded the original 487 estates to 760, which
formed the basis for the time-series plots and modelling analysis.
12 | P a g e
Figures 3a
4
and 3b depict the annual trend of
within the 500- and 1000-metre
areas of MTR lines
5
and two stations, respectively (i.e., Lam Tin and LOHAS Park). While
the revealed trends are not highly significant, it can be observed that at both levels,
housing development (dominated by either public or private housing on a yearly basis)
was roughly more intense within the period of five years before and after a line was
opened. Furthermore, the sheer quantity of housing supply surrounding Lam Tin Station
appears to exceed the average level of all other MTR lines. This may be attributed to the
fact that as a single station, Lam Tin captures a relatively large number of housing units
within its vicinity (refer to Table 1).
<Figure 3 Trend plots of
within the 500-metre and 1000-meter areas of MTR
lines>
To better discern the patterns of housing supply around the MTR network, we divided
the MTR lines and their associated stations into four groups, namely, 1980 and before,
between 1981 and 1990, between 1991 and 2000, and 2001 and after, and we plotted
their respective housing supply trends (Figure 4).
<Figure 4 Trend plots of
within the 500-metre area of MTR lines, by period >
The housing provisions within the 500-metre area of MTR lines built in 1990 and before
appear to have been largely public since Time 0 (marked by a reference line). However,
along some post-2000 MTR lines (e.g., West Rail and Ma On Shan), it is observed that a
more balanced share was achieved between private and public housing development.
Furthermore, more private housing developed within the vicinity of the Kowloon South
Line and the Kwun Tong Line Extension six to ten years before their opening.
Additionally, the land market near the LOHAS Park Station was dominated by private
housing development both before and after it was opened.
We have also examined the housing distribution within the 1000-metre area of MTR
lines
6
. More public housing estates were captured for the period 1980 and before and
the period between 1981 and 1990. Furthermore, echoing the patterns observed in the
500-metre area, private housing development has seen a resurgence, particularly along
the lines opened around 2000. This indicates the possibility that the housing supply
policy associated with the R+P model has become more profit-seeking, in parallel with
these more-recent MTR lines.
Modelling results
4
In Figures 3 and 4, the “Time (in year)” axis is standardised based on the year the MTR line was
opened. For example, “0” represents the open year, and “1” and “-1” represent one year after and
before the opening of the MTR lines, respectively. For example, for the Kwun Tong Line, Time 0 is
1979, and Time -1 and Time 1 are 1978 and 1980, respectively.
5
Since the estate information is collected from the 2011 census, the effects of lines that opened
after 2011—particularly the Island Line Extension (2014), the Kwun Tong Ling Extension (2016)
and the South Island Line (2016)—on new housing estates are largely not captured in the trend
plots. For similar reasons, the effects of the Kowloon Southern Link (2009) may not be fully
captured either, given that it was opened only two years before 2011.
6
Figures for housing distribution within the 1000-metre area of MTR stations by period are not
included in the paper due to limited space, but they can be obtained from the authors upon
request.
13 | P a g e
To examine the effect of MTR accessibility on the location of two types of housing
estates over time, we estimate eight models to cover two different periods: a full period
of 1984-2011 and a more recent period from 1995 to 2011, which can be matched with
available neighbourhood characteristics.
First, we focus on models using all housing estates between 1984 and 2011 (N = 582).
Four models were computed: Model 1 only included accessibility and time dummy
variables with 1984-1988 as the reference group (N = 127); Models 2 and 3 added the
opening of an MTR station within the 500-meter radius of the housing estate and change
in accessibility before and after three or five years; and finally, Model 4 included
regional dummy variables with New Territories (or NT) as the reference group (Table 3).
Neighbourhood variables were, however, not included in the estimation covering 1984-
2011, given that such information is only available in 2001, 2006 and 2011.
An examination of the model fit indices indicates that while statistically significant
results were achieved for all four models, the independent variables explained a small
fraction of the variance of the dependent variable (the Pseudo R2 ranges from 0.04 to
0.11). Two periods, 1989-1997 and 1998-2001, were found to be associated with more
public housing development than the reference group. In addition, the coefficient of
MTR accessibility was negative, suggesting that private housing estates were less likely
to be found in neighbourhoods with higher MTR accessibility, which disagrees with our
hypothesis. Nevertheless, the opening of an MTR station and an increase in accessibility
in a neighbourhood five years before (but not after) a housing estate were positively
associated with the likelihood of a housing estate being privately developed. This
suggests that MTR expansion and the resulting improvement in accessibility may have
attracted private housing development. It is also worth noting that in Model 4, the two
regional dummy variables were found to have statistically significantly negative effects,
suggesting that private housing estates were more likely to be based in NT than HK
Island and Kowloon. This finding helps explain the negative effects of accessibility found
in our models, given that NT, on average, may have a lower level of neighbourhood MTR
accessibility compared to the urban areas (i.e., Kowloon and HK Island).
<Table 3 Modelling results of housing estates 1984-2011>
Next, we move onto models using the housing estates built between 1995 and 2011 (N =
280). The aim is twofold: first, to observe the more-recent impact of MTR accessibility
on housing type; and second, to control for the effects of some of the main
neighbourhood effects. For a given housing estate, its neighbourhood characteristics (i.e.,
population density and the proportion of high-income households) are equal to those of
the corresponding TPU. Housing estates established before 2001, between 2002 and
2006 and between 2007 and 2011 took the variables of 2001, 2006 and 2011,
respectively. Drawing on the same model structures of Models 1 to 4, four models
(Models 5 to 8) were estimated (Table 4) with estates established in 1995-1997 as the
reference group (N = 58).
<Table 4 Modelling results of housing estates 1995-2011>
The four models are all significant and markedly improved in terms of model fit (i.e., all
Pseudo R2 > 0.37) after including neighbourhood variables. Population density was
found to have an insignificant effect on the type of housing development. A higher
proportion of high-income households, on the other hand, has a strong positive effect on
the presence of private housing estates. This finding is largely in line with the fact that
housing prices in upscale neighbourhoods are normally also higher (Bowes and
Ihlanfeldt, 2001; Gibbons and Machin, 2008). The effect revealed for change in
14 | P a g e
accessibility five years before a housing estate was in line with the former four models
in that a positive effect was found to be statistically significant for this variable. MTR
accessibility, however, was largely insignificant, except for a marginally negative effect
in Model 7. This may reflect a closing gap in MTR accessibility between NT and the
urban areas (HK Island and Kowloon) due to the continuing expansion of the MTR
network. Since 2000, quite a few railway lines have been constructed in NT, including
the Ma On Shan, West Rail and Tseung Kwan O Lines, which collectively enhanced this
area’s MTR accessibility. Lastly, the effects of the regional dummy variables were
insignificant in Models 5-8, suggesting that there might have been less land in NT that is
readily available for housing development.
To corroborate the above findings, we mapped out the locations of the housing estates
of the two periods, namely, 1984-1994 as a full period and 1995-2011 as a more recent
period (Figures 5 and 6). From Figure 5, it can be observed that more private housing
estates appeared in more-remote areas (NT) than in central areas (HK Island and
Kowloon), although this disparity has declined since 1995 (Figure 6). An examination of
the standardised MTR accessibility in 1995 and 2011 indicates that the main
improvement in MTR accessibility between the two time points indeed occurred in NT,
particularly along the MTR lines (e.g., Ma On Shan Line and West Rail Line). Both
findings affirm our speculation regarding the location choice of private housing
development, namely, that compared to public housing, it is more concentrated in the
suburban areas, particularly those associated with significant improvement in MTR
accessibility.
<Figure 5 Location of housing estates 1984-1994>
<Figure 6 Location of housing estates 1995-2011>
CONCLUSION
Transit-induced injustice has been highlighted as an emerging issue, although empirical
evidence in TOD has been scant (Grube-Cavers and Patterson, 2014; Revington, 2015).
Hence, efforts should be made to assess and monitor potentially imbalanced housing
markets in relation to TOD projects. Doing so may help to avoid the social-spatial
marginalisation of certain socio-demographic groups in the process of urban
development (Kahn, 2007; Grube-Cavers and Patterson, 2014). To this end, we
investigated the development of the MTR system in Hong Kong and its impact on the
opening of new public and private housing estates.
Although Hong Kong’s MTR has been acclaimed for its R+P business model, there has
been concern over whether this model has prioritised the development of private
housing estates within the precincts of the railway services, thereby potentially
marginalising public housing and their residents. To shed light on this issue, we
conducted a series of visual and statistical analyses on the location choice of private and
public housing estates and its association with MTR development. The key findings and
their implications are summarised as follows.
First, longitudinal trends of public versus private housing development along individual
MTR lines were plotted. In part agreeing with previous research (Bowes and Ihlanfeldt,
2001; Sun et al., 2015), the plots point towards a capitalisation of both anticipated and
realised benefits of MTR expansion into the housing market. For most lines, this
capitalisation effect appears to be more pronounced five years before and after the
opening of a new line. Furthermore, while in the earlier years more public housing
15 | P a g e
estates were opened in the precincts of MTR stations, the numbers are more balanced
more recently (i.e.., since 2000). In other words, these prime locations are no longer
dominated by (or reserved for) public housing, which may have given the long-term
residents of Hong Kong the impression that more private housing estates have taken up
the prime locations near MTR. Furthermore, it is noteworthy that around 2003, the
government reduced the provision of public housing under the pressure exerted by
private developers (Chen and Pun, 2007). This may also have contributed to the
observed reduction of public housing in the precincts of the MTR lines opened after
2000.
Next, a series of logistic regression models were estimated to quantify the impacts of
MTR accessibility on the type of housing development. Largely contrary to our
hypothesis, regional MTR accessibility was found have either an insignificant or a
negative effect on the presence of a private housing estate. This indicates that locations
associated with higher MTR accessibility did not necessarily invite more private housing
development. It was also found that compared to public housing estates, private housing
estates were slightly more concentrated in the suburban (NT) areas, particularly before
1995. This may have neutralised, to some extent, the expected importance of MTR
accessibility in attracting private housing development. Nevertheless, the accessibility
improvement resulting from the opening of new MTR stations and the expansion of MTR
lines over time does appear to have consistently attracted private housing development
regardless of specific regions. This provides some evidence for the speculative
behaviour of private developers. In other words, private developers may tend to stock
up land reserves with relatively low accessibility and delay development until a
significant change in transport accessibility has been realised through government
infrastructure projects.
Combining the time-series plots with the modelling results, we can conclude that
although more land adjacent to the MTR network has been allocated to public housing,
there appears to have been a shift in housing supply policy associated with MTR
development, which may have given higher priority to private-sector investment (to the
detriment of public housing development) over the past two decades. Given this
scenario, although our findings do not necessarily suggest the existence of severe spatial
injustice under the TOD model, caution should be taken when planning for future urban
development should this trend continue. Strategically, over-reliance on a profit-seeking
paradigm (particularly since its Initial Public Offering in 2000) to operate and manage
MTR-adjacent development should be avoided, given its potential to result in a less
equitable housing market for those who cannot afford private dwellings. More
particularly, considering the continuing enhancement of MTR service and accessibility,
especially around the New Territories region, land prices are also likely to keep rising.
As such, the involvement of private developers in the land market adjacent to the newly
planned MTR stations should be controlled at a reasonable level so that public housing
development will not be outbid in these prime locations. Therefore, we would
recommend that reserving sufficient land for public housing should be explicitly
incorporated into the MTR expansion plan, both in New Territories and in the urban
areas. After all, MTR should strike a balance between efficiency and equity as a transit
operator-cum-estate developer.
To conclude, our study contributes to the existing debate on transit-induced injustice
through an empirical investigation in an Asian transit city. Our findings show that when
the planning, building and operating of a public transport system is closely link to land
development under the influence of a profit seeking model, its network expansion can
have strong implications for the supply of the local housing market and, hence, the
residential locations of different social strata. Our research adds value to the current
16 | P a g e
understanding of the interplay between transport and land use and to the latest
inquiries into the socio-spatial impact of transit-oriented development. More research is
needed to further probe into the issue of transit-induced injustice.
Acknowledgements
This research was supported by the CUHK Social Science Collaborative Research Fund
(SS14461).
REFERENCES
BARTHOLOMEW, K., and EWING, R. (2011), ‘Hedonic price effects of pedestrian-and
transit-oriented development’, Journal of Planning Literature, 26, 18-34.
BOWES, D. R., and IHLANFELDT, K. R. (2001), ‘Identifying the impacts of rail transit
stations on residential property values’, Journal of Urban Economics, 50, 1-25.
CENSUS AND STATISTICS DEPARTMENT (2012), Major Housing Estates,
http://www.census2011.gov.hk/en/major-housing-estates.html (accessed 1
February 2017).
CENSUS AND STATISTICS DEPARTMENT (2016), Population,
http://www.censtatd.gov.hk/hkstat/sub/so20.jsp (accessed 1 August 2017).
CERVERO, R. (1998), The Transit Metropolis: A Global Inquiry, Washington, D.C., Island
Press.
CERVERO, R. (2006), Public Transport and Sustainable Urbanism: Global Lesson, Berkeley,
University of California Transportation Center.
CERVERO, R. (2009), ‘Transport infrastructure and global competitiveness: balancing
mobility and livability’, ANNALS of the American Academy of Political and Social
Science, 626, 210-25.
CERVERO, R. and DUNCAN, M. (2002), ‘Transit's value-added effects: light and
commuter rail services and commercial land values’, Transportation Research
Record: Journal of the Transportation Research Board, 1805, 8-15.
CERVERO, R., and MURAKAMI, J. (2009), ‘Rail and property development in Hong Kong:
experiences and extensions’, Urban Studies, 46, 2019-43.
CHEN, Y.C., and PUN, N. (2007). ‘Neoliberalization and privatization in Hong Kong after
the 1997 financial crisis’, China Review, 7, 65-92.
CHIU, R. L. (2007), ‘Planning, land and affordable housing in Hong Kong’, Housing Studies,
22, 63-81.
CIVIL ENGINEERING AND DEVELOPMENT DEPARTMENT, HK (2016), New Towns, New
Development Areas and Urban Developments,
https://www.gov.hk/en/about/abouthk/factsheets/docs/towns%26urban_devel
opments.pdf (accessed 1 August 2017).
CURTIS, C., RENNE, J. L. and BERTOLINI, L. (eds) (2009), Transit Oriented Development:
Making It Happen, Surrey, Ashgate.
DELMELLE, E. C., and CASAS, I. (2012), ‘Evaluating the spatial equity of bus rapid transit-
based accessibility patterns in a developing country: the case of Cali, Colombia’,
Transport Policy, 20, 36-46.
DU, H., and MULLEY, C. (2006), ‘Relationship between transport accessibility and land
value: local model approach with geographically weighted regression’,
Transportation Research Record: Journal of the Transportation Research Board,
1977, 197-205.
DUNCAN, M. (2011), ‘The impact of transit-oriented development on housing prices in
San Diego, CA’, Urban Studies, 48, 101-27.
DUNWORTH, T., and SAIGER, A. J. (1994). Drugs and Crime in Public Housing: A Three-
city Analysis, Washington, D.C., US Department of Justice, Office of Justice Programs,
National Institute of Justice.
17 | P a g e
FOTH, N., MANAUGH, K., and EL-GENEIDY, A. M. (2013), ‘Towards equitable transit:
examining transit accessibility and social need in Toronto, Canada, 1996–2006’,
Journal of Transport Geography, 29, 1-10.
FRIEDMAN, L. M. (1966), ‘Public housing and the poor: an overview’, California Law
Review, 54, 642-669.
GEURS, K. T., BOON, W., and VAN WEE, B. (2009), ‘Social impacts of transport: literature
review and the state of the practice of transport appraisal in the Netherlands and
the United Kingdom’, Transport Reviews, 29, 69-90.
GIBBONS, S., and MACHIN, S. (2008), ‘Valuing school quality, better transport, and lower
crime: evidence from house prices’, Oxford Review of Economic Policy, 24, 99-119.
GIULIANO, G., REDFEARN, C., AGARWAL, A., and HE, S. (2012), ‘Network accessibility
and employment centres’, Urban Studies, 49, 77-95.
GRUBE-CAVERS, A. and PATTERSON, Z. (2015), ‘Urban rapid rail transit and
gentrification in Canadian urban centres: a survival analysis approach’, Urban
Studies, 52, 178-94.
HONG, Y. H. (1998), ‘Transaction costs of allocating increased land value under public
leasehold systems: Hong Kong’, Urban Studies, 35, 1577-95.
KAHN, M. E. (2007), ‘Gentrification trends in new transit‐oriented communities:
evidence from 14 cities that expanded and built rail transit systems’, Real Estate
Economics, 35, 155-82.
KAY, A. I., NOLAND, R. B. and DiPETRILLO, S. (2014), ‘Residential property valuations
near transit stations with transit-oriented development’, Journal of Transport
Geography, 39, 131-40.
KNOWLES, R. D. (2012), ‘Transit oriented development in Copenhagen, Denmark: from
the finger plan to Ørestad’, Journal of Transport Geography, 22, 251-61.
LEE, S. L., YUAN, L. L. and POH, T. K. (1993), ‘Shelter for all: Singapore's strategy for full
home ownership by the year 2000’, Habitat International, 17, 85-102.
LEE, J. and YIP, NM. (2006), ‘Public housing and family life in East Asia: housing history
and social change in Hong Kong, 1953-1990’, Journal of Family History, 31, 66-82.
LIU, S. and ZHU, X. (2004), ‘Accessibility analyst: an integrated GIS tool for accessibility
analysis in urban transportation planning’, Environment and Planning B: Planning
and Design, 31, 105-24.
MATHUR, S. and FERRELL, C. (2013), ‘Measuring the impact of sub-urban transit-
oriented developments on single-family home values’, Transportation Research
Part A: Policy and Practice, 47, 42-55.
MOHAMMAD, S. I., GRAHAM, D. J., MELO, P. C. and ANDERSON, R. J. (2013), ‘A meta-
analysis of the impact of rail projects on land and property values’, Transportation
Research Part A: Policy and Practice, 50, 158-70.
MTR. (2017), Business Overview,
http://www.mtr.com.hk/archive/corporate/en/publications/images/business_ov
erview_e.pdf (accessed 1 September 2017).
MTR. (2015), Financial Performance,
http://www.mtr.com.hk/en/corporate/sustainability/2014rpt/finan-
business.php (accessed 1 August 2017).
MTR Corporation Limited (2016), Annual Report 2015, Hong Kong, MTR.
OAKLEY, D. and BURCHFIELD, K. (2009), ‘Out of the projects, still in the hood: the spatial
constraints on public-housing residents’ relocation in Chicago’, Journal of Urban
Affairs, 31, 589-614.
OLARU, D., SMITH, B. and TAPLIN, J. H. (2011), ‘Residential location and transit-oriented
development in a new rail corridor’, Transportation Research Part A: Policy and
Practice, 45, 219-37.
RATNER, K. A. and GOETZ, A. R. (2013), ‘The reshaping of land use and urban form in
Denver through transit-oriented development’, Cities, 30, 31-46.
18 | P a g e
REVINGTON, N. (2015), ‘Gentrification, transit, and land use: moving beyond
neoclassical theory’, Geography Compass, 9, 152-63.
ROHE, W. M. and BURBY, R. J. (1988), ‘Fear of crime in public housing’, Environment and
Behavior, 20, 700-20.SUN, W., ZHENG, S. and WANG, R. (2015), ‘The capitalization
of subway access in home value: a repeat-rentals model with supply constraints in
Beijing’, Transportation Research Part A: Policy and Practice, 80, 104-15.
SARD, B., and FISCHER, W. (2008). Preserving safe, high quality public housing should be
a priority of federal housing policy. Center on Budget and Policy Priorities, 1, 36.
TANG, B. S., CHIANG, Y. H., BALDWIN, A. N. and YEUNG, C. W. (2004), Study of the
Integrated Rail-property Development Model in Hong Kong, Hong Kong, Research
Center for Construction and Real Estate Economics, the Hong Kong Polytechnic
University.
TANG, S. and LO, H. K. (2008), ‘The impact of public transport policy on the viability and
sustainability of mass railway transit: the Hong Kong experience’, Transportation
Research Part A: Policy and Practice, 42, 563-76.
TRANSPORT DEPARTMENT, HK (1993), Travel Characteristics Survey Executive
Summary, Hong Kong, Transport Department.
TRANSPORT DEPARTMENT, HK (2003), Travel Characteristics Survey 2002 Final Report,
Hong Kong, Transport Department.
TRANSPORT DEPARTMENT, HK (2014), Travel Characteristics Survey 2011 Final Report,
Hong Kong, Transport Department.
UN-HABITAT (2004), ‘The challenge of slums: global report on human settlements 2003’,
Management of Environmental Quality: An International Journal, 15, 337-38.
WAKELY, P. (2014), ‘Urban public housing strategies in developing countries: whence
and whither paradigm, policies, programmes and projects’, DPU60 Working Paper
Series: Reflections NO. 163/60.
WAKELY, P. and RILEY, E. (2011), The Case for Incremental Housing: Policy Research and
Working Paper Series No. 1, Washington, DC, Cities Alliance.
WON, S., CHO, S. E. and KIM, S. (2015), ‘The neighborhood effects of new road
infrastructure: transformation of urban settlements and resident's socioeconomic
characteristics in Danang, Vietnam’, Habitat International, 50, 169-79.
19 | P a g e
Appendix
List of main independent variables
Variable
Description
PopDen
Population per square kilometre
HiIncHou
Proportion of households with a monthly income level of
HK$40,000 or more
Accessibility
Standardised MTR accessibility ψ
Acc3yBef
Change in standardised MTR accessibility ψ three years before ζ
Acc3yNex
Change in standardised MTR accessibility ψ three years after ζ
Acc5yPas
Change in standardised MTR accessibility ψ five years before ζ
Acc5yNex
Change in standardised MTR accessibility ψ five years after ζ
Sta3yPas
Opening of an MTR station within 500 m in the past three years ζ
Sta3yNex
Opening of an MTR station within 500 m in the next three years ζ
Sta5yPas
Opening of an MTR station within 500 m in the past five years ζ
Sta5yNex
Opening of an MTR station within 500 m in the next five years ζ
ψ at street block level
ζ benchmark temporal point: opening of the housing estate
20 | P a g e
Table 1 Description of MTR Lines
Line
Opening
year
Num. of
stations
Within 500-metre radius
Within 1,000-metre radius
Total housing
units †
Public
housing (%)
Private
housing (%)
Total housing
units †
Public
housing (%)
Private
housing (%)
Kwun Tong Line
1979
11
98,791
83
17
243,791
86
14
East Rail Line
1982
11
47,160
65
35
171,458
68
32
Tsuen Wan Line
1982
14
84,223
73
27
171,234
71
29
Island Line
1985
12
76,522
43
57
95,454
52
48
Lam Tin Station
1989
1
17,803
67
33
55,846
78
22
Tung Chung Line
1998
5
30,511
35
65
72,721
40
60
Tseung Kwan O Line
2002
5
97,491
63
37
131,975
71
29
West Rail Line
2003
8
41,847
84
16
148,725
63
37
Ma On Shan Line
2004
8
58,066
60
40
110,597
64
36
Kowloon Southern Link*
2009
2
2,560
0
100
4,975
31
69
LOHAS Park Station
2009
1
8,016
0
100
8,016
0
100
Island Line Extension
2014
3
5,729
49
51
6,877
49
51
Kwun Tong Line Extension
2016
4
45,198
21
79
45,198
21
79
South Island Line
2016
4
19,412
49
51
30,376
58
42
*The Kowloon Southern Link is a segment of the West Rail Link, linking Nam Cheong Station of the West Rail Line with Hung Hom Station of the East Rail Line. This
link itself includes two stations, Austin and East Tsim Sha Tsui.
† up to the 2011 Census
21 | P a g e
Table 2 Descriptive statistics of housing estates
Public housing estate
Private housing estate
Housing and population
statistics
No. of housing units
1,015,163
439,992
No. of households
1,105,174
390, 517
Population
3,323,123
1,166,675
Average of median household
monthly income (in HK$)
18,204
44,567
Household weighted average of
median household monthly
income (in HK$)
16,310
43,047
Walking time to the nearest
MTR station (min)
Mean
16.5
17.5
Population weighed mean
16.8
18.9
Median
12.5
10.2
Std. deviation
13.5
19.5
22 | P a g e
Table 3 Modelling results of housing estates 1984-2011
Model 1
Model 2
Model 3
Model 4
Variable
coef.
coef.
coef.
coef.
Cons
0.33*
0.16
-0.02
0.2
1989-1997
-0.64***
-0.55**
-0.44*
-0.4*
1998-2001
-0.89***
-0.98***
-0.83***
-0.76**
2002
0.54
0.55
0.4
0.54
2003
0.56
0.52
0.67
0.75
2004-2008
-0.15
-0.07
-0.01
0.05
2009-2011
-0.76
-0.76
-0.78
-0.68
Accessibility
-0.38***
-0.5***
-0.65***
-0.44***
Sta3yearBef
1.48***
Sta3yearNex
0.76
Sta5yearBef
1.33***
1.38***
Sta5yearNex
0.77
0.76
Acc3yearBef
0.39
Acc3yearNex
-0.23
Acc5yearBef
0.89***
0.67**
Acc5yearNex
-0.42
-0.46
HK Island
-0.6*
Kowloon
-0.64**
Model summary
2
33.88***
56.28***
78.91***
85.74***
Pseudo R2
0.042
0.0698
0.0978
0.1063
Note: *** p<0.01; ** p<0.05; * p<0.1
23 | P a g e
Table 4 Modelling results of housing estates 1995-2011
Model 5
Model 6
Model 7
Model 8
Variable
coef.
coef.
coef.
coef.
Cons
-4.1***
-4.41***
-4.56***
-4.31***
PopDen
-0.01*
-0.008
-0.005
-0.001
HiIncHou
15.9***
16.57***
16.53***
16.27***
1998-2001
1.62***
1.46***
1.4***
1.41***
2002
2.83***
2.47***
2.11**
2.27***
2003
3.05***
3.04***
3.05***
3.07***
2004-2008
1.83***
1.81***
1.63***
1.64***
2009-2011
-0.58
-0.52
-0.72
-0.73
Accessibility
-0.2
-0.34
-0.49*
-0.24
Sta3yearBef
1.07
Sta3yearNex
-0.46
Sta5yearBef
0.93
0.9
Sta5yearNex
-0.15
-0.18
Acc3yearBef
1.01
Acc3yearNex
0.5
Acc5yearBef
1.3**
1.04*
Acc5yearNex
0.18
0.18
HK Island
-0.64
Kowloon
-0.74
Model summary
2
144.39***
153.13***
158.09***
160.14***
Pseudo R2
0.3747
0.3974
0.4102
0.4156
Note: *** p<0.01; ** p<0.05; * p<0.1
24 | P a g e
Figure 1 MTR and the R+P model
Source: MTR (2015)
25 | P a g e
Figure 2 Location of housing estates
26 | P a g e
Figure 3 Trend plots of
within the 500-metre and 1000-metre areas of MTR lines
27 | P a g e
Figure 4 Trend plots of
within the 500-metre area of MTR lines, by period
28 | P a g e
Figure 5 Location of housing estates 1984-1994
29 | P a g e
Figure 6 Location of housing estates 1995-2011
